| Summary: | 碩士 === 國立中正大學 === 電機工程研究所 === 105 === With the help of rapid advancing electronic technology, shopping stores have gradually been transformed from a passive exhibition space to a commercial organism that can actively collect consumers' interests and accordingly push for sale. As far as collection is concerned, there have been plenty of research works that integrate many sensors for the job. This thesis explores the potential of using only image processing techniques on surveillance video to accomplish such collection. Two behaviors, namely, grabbing and watching, are proposed to reveal the target consumer's shopping interest in the store. Algorithms are also devised to identify grabbing and to discern watching as well as the watching direction. Experimental results in a prototype store shows that the proposed techniques have already exhibited great potential in collecting consumers' interests.
|
|---|
